A bank's cryptography is spread across hundreds of systems it did not write: card HSMs (hardware security modules), payment switches, core banking platforms, branch routers and mobile apps. Most of the public-key part of that estate (RSA, elliptic-curve and Diffie-Hellman key exchange, signatures and certificates) will need to move to post-quantum algorithms. This page sets out what Indian regulators have actually said so far, where the cryptography usually sits in a bank, and what to do in the next two quarters.
What your regulators expect
RBI has not yet issued a binding direction that requires post-quantum cryptography (PQC) or a cryptographic inventory. It has set up an expert committee whose terms of reference include exactly that. The binding rules that apply today are the general cryptographic-strength clauses in RBI's 2026 directions.
RBI: Q-SAFE expert committee (25 May 2026)
On 25 May 2026 RBI announced an Expert Committee on a Quantum Secure and Adaptive Financial Ecosystem (Q-SAFE)[1]. The committee's convener is a professor from IIT Madras. Its members include representatives of DST, MeitY, SBI, NPCI and DSCI, and its member-secretary comes from RBI's FinTech Department. Its terms of reference include assessing the financial sector's cryptographic inventory using a CBOM (cryptographic bill of materials), evaluating crypto agility and vulnerable systems, assessing industry readiness, and recommending a roadmap and framework. The report is due within six months of the committee's first meeting. Treat it as the most likely source of any future RBI PQC requirement. The CBOM wording in its terms of reference is a good hint of what banks may be asked for.
RBI: 2026 Cyber Directions and Digital Payment Security Controls (31 July 2026). These are binding.
The Reserve Bank of India (Commercial Banks – Cybersecurity, Technology: Risk, Resilience and Assurance Framework) Directions, 2026 took effect on issue. They apply to commercial banks other than small finance banks, payments banks and local area banks, and they repeal the earlier cybersecurity framework and IT governance instructions for those banks[2]. Three clauses matter here:
- Para 140: key lengths, algorithms, cipher suites and protocols "shall be strong", and the bank "shall adopt internationally accepted and published standards that are not deprecated / demonstrated to be insecure / vulnerable".
- Para 47: an up-to-date inventory of information assets, showing their business criticality.
- Para 37: avoid outdated and unsupported hardware or software, and monitor end-of-support dates.
The companion Digital Payment Security Controls Directions, 2026 (same date, same scope) go further for payment channels[3]. Para 24 lists what must be strong: key lengths for encryption and hashing; algorithms for encryption, signing, key exchange, message digests and random number generation; cipher suites; digital certificates; and protocols. Para 69(5) sets HSM controls, including secure management of Local Master Keys (LMKs) and controls to stop PIN blocks being output in weaker formats.
Neither direction mentions quantum computing. But the "not deprecated" test refers to international standards, and those standards now carry dates. NIST IR 8547 (final, October 2026) deprecates quantum-vulnerable public-key algorithms after 2030 and disallows them after 2035[9]. This is our reading, not RBI's: the same paragraph that is satisfied by RSA-2048 today will be harder to argue for once those dates arrive.
RBI's leadership has also spoken about the risk. On 11 September 2026 Deputy Governor Shirish Chandra Murmu said that payment infrastructure "has long technology cycles and deeply embedded cryptographic dependencies, and migration to new standards may take years". He also named the "harvest now, decrypt later" risk[4]. That is a speech, not a direction.
CERT-In v2.0 (9 July 2025). This is technical guidance.
CERT-In's Technical Guidelines on SBOM, QBOM & CBOM, AIBOM and HBOM say that government, public sector and essential services organisations "shall require" a CBOM in "all related procurements, developments, and integrations" (§8.4.1.1). They also recommend tiered vendor PQC contract terms (§8.5.4)[5]. The wording is mandatory in tone, but how it binds a particular bank depends on its regulator and its contracts.
DST / National Quantum Mission roadmap (May 2026). This is advisory.
The DST task-force report lists Banking, Financial Services and Insurance (BFSI) among the critical information infrastructure (CII) sectors on an accelerated timeline[6]:
- foundations by 31 December 2027;
- high-priority systems migrated by 31 December 2028;
- full PQC adoption by 31 December 2029.
It also advises asking vendors for CBOMs from FY 2026–27 and making them mandatory in procurement from FY 2027–28. It asks for the report to be sent to regulators, RBI among them, so they can issue sector-specific guidance. Until RBI does so, these dates are a planning baseline, not a compliance deadline.
The Reserve Bank Innovation Hub's 2024 whitepaper on PQC in Indian banking recommends starting with a cryptographic inventory covering applications, infrastructure and networks[7]. It is research guidance, not regulation.
The procurement signal: SBI RFP 1506
Large banks are already writing these requirements into contracts. SBI's software RFP ref. 1506, dated 12 October 2026, requires a complete SBOM (Annexure R). Wherever applicable, it also requires a QBOM, CBOM, AIBOM and HBOM (Annexures S–V). These must be kept current and submitted at onboarding and on any material change[8]. Expect vendors to see similar clauses from other banks.
Where the cryptography sits in a bank
The table below shows typical patterns, not findings about any particular bank. Priorities assume a large commercial bank and should be re-ranked against your own data shelf life and vendor roadmaps.
| System | Cryptography typically in use | Why it matters for quantum migration | Priority |
|---|---|---|---|
| Internet and mobile banking | TLS with RSA or ECDSA certificates and ECDHE key exchange; app signing; device-binding keys | Internet-facing sessions carry credentials and account data that can be recorded now and decrypted later | P0: largest exposure to recorded traffic |
| UPI / IMPS integrations and payment switches | Mutual TLS, PKI-signed and encrypted API messages, HSM-held keys for PIN handling | Changes depend on NPCI and every counterparty moving together; you cannot migrate alone | P0 for inventory; migration paced with the ecosystem |
| Card processing and HSMs | Symmetric keys under LMKs, PIN blocks, public-key certificates on chip cards and in key exchange | Symmetric parts mostly need adequate key length; public-key parts and HSM firmware have long replacement cycles | P1: and P0 for any HSM purchase now |
| SWIFT and interbank messaging | PKI certificates, HSM-held signing keys, TLS or VPN links | Signatures give long-lived non-repudiation; the migration path is set by the network operator | P1: track the operator's roadmap |
| Core banking and internal APIs | TLS between tiers, database and backup encryption, service-account certificates, vendor libraries | Largest blast radius and heavy vendor dependence; you need the vendor's CBOM to see inside | P1: drive through vendor CBOMs |
| Branch and corporate VPNs | IPsec/IKE with Diffie-Hellman or ECDH, RSA device certificates | Long-lived traffic can be harvested; appliances need firmware that supports PQC | P1: often tied to hardware refresh |
| Code signing and software updates | RSA or ECDSA signatures on apps, ATM software, internal releases | A broken signing key lets an attacker push trusted malware; roots of trust last many years | P1: plan new roots early |
| ATM network | TLS or VPN to the switch, remote key loading, terminal master keys | Long terminal lifecycles; remote key loading uses public-key cryptography | P2: P1 where remote key loading is used |
| Data at rest and backups | AES, with keys often wrapped under RSA in a key manager | Long retention; AES-256 holds up, but RSA key wrapping does not | P2: check key wrapping and AES-128 use |
Where to start
- Build a CBOM for the P0 surface. Start outside-in with a free public TLS scan of your banking domains. Then inventory internet banking, mobile and UPI integration endpoints using the approach in how to find weak cryptography and the CBOM guide. Q-SAFE's terms of reference make a CBOM the likely unit of evidence.
- Write CBOM clauses into every new contract this quarter. SBI's RFP 1506 and DST's FY 2026–27 advice point the same way. Use the PQC procurement requirements guide. For HSMs, see HSM procurement for PQC. Check what vendors send against how to validate a CBOM.
- Rank systems by data shelf life, not only by criticality. Use the Mosca inequality to decide which channels face harvest-now-decrypt-later exposure first.
- Put DST's CII dates in front of the board. Map your inventory to the 2027 / 2028 / 2029 milestones on the India roadmap, so the plan is ready when RBI's guidance arrives.
- Benchmark yourself. See where peers stand on the India BFSI PQC readiness index.
How KavachQ helps
- Find it. KavachQ Discover inventories the cryptography across internet banking, APIs, keystores, code and configuration, and gives every asset a T1–T4 quantum-risk tier. HSM and payment-system configurations are parsed from exports you provide; KavachQ does not connect to your HSMs.
- Map it to RBI. The regulatory map covers RBI alongside DST/NQM, SEBI, CERT-In and IRDAI, and Kavach Copilot answers questions from the regulators' verbatim text.
- Prove it. The inventory is a CycloneDX 1.7 CBOM signed with ML-DSA-65 that an auditor can verify offline against the signing key's published fingerprint: evidence in the CBOM form that Q-SAFE's terms of reference name.
- Fix it safely. Migrate plans P0/P1/P2 work against DST/NQM dates and can change TLS terminators (nginx, HAProxy and Apache, plus ACME certificate renewal) after a named approver signs off, with a tested rollback.
Software vendors selling to banks can use KavachQ Declare for SBI-style BOM annexures. See how India's banks measure up in the BFSI Readiness Index.